Most companies decide what to write by asking what to rank for. That's the wrong order. The keyword tells you what people are searching; it doesn't tell you whether you have anything to say about it. And in 2026, an article that exists because a keyword exists is an article AI can write faster and cheaper than you can.
The title is where you find out if the article should exist at all.
The keyword trap
For a long time, the workflow looked reasonable. Pick a keyword. Cluster the why/what/how/where questions around it. Assign the brief. Ship the post. This was a real upgrade from keyword stuffing, and for years it was enough — generic content ranked because there wasn't much competition for decent.
That window closed. AI raised the floor on what counts as decent, and search traffic is harder to earn than it's been in a decade. The question-cluster method still produces articles, but the articles it produces are exactly the ones a language model can replicate in seconds. They define a concept. They answer a literal question. They have no argument.
The problem isn't the keyword. The problem is that the keyword is doing all the thinking.
What a title is supposed to do
A title is the compressed version of the argument. If you can't write a title that's interesting, it's because there isn't an interesting article underneath it yet. The title isn't the last thing you write. It's the first test the idea has to pass.
This is not a clickbait argument. Clickbait titles promise more than the article delivers. Good titles promise exactly what the article delivers, and what the article delivers is worth promising. "10 ways to improve your content strategy" fails this test because it isn't promising anything specific. All it's doing is offering a familiar format.
A working title states a position. Your blog title is your content strategy. Most B2B writing is paragraphs that gave up. The em-dash is not a thinking tool. You may disagree with any of these. That's the point.
A title someone can disagree with is a title with an article inside it.
Build the filter into the calendar
Most content calendars are lists of topics. They should be lists of arguments.
The fix is small and structural: between "topic penciled in" and "writer assigned," add a step where someone has to write a one-sentence thesis. Not a brief. Not an outline. A sentence that takes a position. If the thesis is interesting, the article gets written. If the thesis is a restatement of the keyword, the slot goes back on the calendar and the topic gets reworked or killed.
This sounds like overhead. It is the opposite of overhead. The expensive thing is writing 1,500 words on a topic that turns out to have nothing under it. The cheap thing is finding that out at the sentence stage.
You will kill more topics than you expect. That is the system working.
What this changes
When titles drive the calendar, three things shift. Writers stop producing articles they don't believe in, because every piece has to clear a thesis bar to get assigned. Editors stop rescuing drafts in revision because the drafts arrive with something to defend. And the blog starts to sound like it was written by people with positions, not by people fulfilling a quota.
None of that requires more resources. It requires moving the hardest decision — is this worth saying? — to the front of the process instead of the end.
The topic is the first idea you create when writing an article. If you aim at nothing, you'll hit it every time.